Currently E-learning is everywhere in the education field. The performance or outcome of the student in E-learning depends on the assessment process happened through online. One of the components in the online assessment is the descriptive answer evaluation system. Most of the assessment process focusses on only Multiple-choice questions and it is easy to evaluate those type of questions through online. When it comes to descriptive answers, evaluation is the challenging task and requires more effort to award the score to a student. The answer given by the student is being compared with desired answer through the Natural Language Processing techniques and the overall Relevance score is being calculated using machine learning. This research work focuses towards automatic awarding of scores to the answers and the result is compared with examiners.